Missing dependencies for v0.6.2 features (sentence-transformers, numpy)

[mimosel]

Summary

The v0.6.2 release introduces "Mid-Stream Semantic Memory" and "Plan Mode Integration" as headline features, but the required dependencies (sentence-transformers, numpy) are not declared in requirements.txt or pyproject.toml. The features silently fall back to keyword matching without any warning.

Environment

• ELF Version: v0.6.2 (commit 4a05771)
• OS: Windows 11
• Python: 3.11
• Discovery Date: 2026-01-29

Problem Description

Missing Dependencies

Dependency	Required By	Status in requirements.txt	Status in pyproject.toml
sentence-transformers	semantic_search.py, semantic-memory.py	MISSING	MISSING
numpy	semantic_search.py (vector operations)	MISSING	MISSING
openai (optional)	semantic_search.py (fallback embeddings)	MISSING	MISSING

Affected Files

src/query/semantic_search.py (lines 27-30, 23)
.hooks-templates/PreToolUse/semantic-memory.py (line 26)
src/hooks/learning-loop/pre_tool_semantic_memory.py (line 26)

Code Evidence

The code uses try/except for graceful degradation:

# src/query/semantic_search.py
try:
    from sentence_transformers import SentenceTransformer
    SENTENCE_TRANSFORMERS_AVAILABLE = True
except ImportError:
    SENTENCE_TRANSFORMERS_AVAILABLE = False

This means:

• No error is raised
• Features silently fall back to keyword matching
• Users think they have semantic search, but they don't

Impact

1. Misleading release notes: v0.6.2 advertises "Mid-Stream Semantic Memory" but it doesn't work out of the box
2. Silent degradation: No warning that semantic features are disabled
3. Installation incomplete: pip install -r requirements.txt does not provide full functionality
4. User confusion: "Why isn't semantic search finding relevant heuristics?"

Steps to Reproduce

1. Fresh install of ELF v0.6.2
2. Run pip install -r requirements.txt
3. Check for sentence-transformers:

   python -c "from sentence_transformers import SentenceTransformer; print('OK')"
   # Result: ModuleNotFoundError

4. The semantic memory hook will run but fall back to keyword matching

Expected Behavior

Either:

1. Option A: Dependencies should be in requirements.txt so features work after standard install
2. Option B: Dependencies should be in pyproject.toml under [project.optional-dependencies] with clear documentation
3. Option C: A warning should be logged when semantic features are disabled due to missing dependencies

Suggested Fix

Option A: Add to requirements.txt

# Semantic search (for mid-stream memory)
sentence-transformers>=2.3.0
numpy>=1.24.0

Option B: Add as optional dependency in pyproject.toml

[project.optional-dependencies]
semantic = [
    "sentence-transformers>=2.3.0",
    "numpy>=1.24.0",
]

With install command: pip install emergent-learning-framework[semantic]

Option C: Add warning when disabled

if not SENTENCE_TRANSFORMERS_AVAILABLE:
    import warnings
    warnings.warn(
        "sentence-transformers not installed. Semantic search disabled. "
        "Install with: pip install sentence-transformers",
        UserWarning
    )

Current Workaround

Manual installation:

pip install sentence-transformers numpy

Note: sentence-transformers is ~500MB including PyTorch dependencies.

Additional Context

The documentation in docs/mid-stream-semantic-memory.md does mention the dependency on line 145-146, but:

1. This file is not prominently linked from README
2. The release notes don't mention additional dependencies
3. Users following standard install procedures won't see this

[Spacehunterz]

Migrated to Ollama (no sentence-transformers needed)

• Graceful degradation when Ollama unavailable
• Install scripts now help users set up Ollama
• v0.6.3 has fixes